Target Stream Research Articles

Small‐Scale Recovery of a Native White‐Spotted Charr Population by Suppression of Non‐Native Brown Trout

ABSTRACTIf eradication is not feasible, suppressing a population size of non‐native species by removal is an alternative option for the conservation of native species that are impacted by non‐native species invasions. This study reports on the suppression of non‐native brown trout for the conservation of native white‐spotted charr in an upstream of Monbetsu Stream, Hokkaido, northern Japan. Brown trout removals were conducted from 2014 to 2023 by electrofishing. The population size of brown trout was dramatically reduced during the first three years of removals, and the ratio of white‐spotted charr to brown trout abundance gradually increased. However, the population size of white‐spotted charr did not fully recover, which can be explained by the openness of the system and likely also by the influence of interfering factors observed during the study, such as changes in stream morphology due to flooding and suspected illegal fish stocking. The abundance of young‐of‐the‐year brown trout also tended to be reduced when numerous mature females had been removed in previous years. This study suggests that suppressing non‐native fish species, even on a limited scale, can contribute to the conservation of native species. However, when preventing migration from downstream into the target stream reach is unfeasible, ongoing suppression of non‐native species becomes essential to ensure the conservation of native species in the area.

TSFormer: Tracking Structure Transformer for Image Inpainting

Recent studies have shown that image structure can significantly facilitate image inpainting. However, current approaches mostly explore structure prior without considering its guidance to texture reconstruction, leading to performance degradation. To solve this issue, we propose a two-stream Tracking Structure Transformer (TSFormer), including structure target stream and image completion stream, to capture the synchronous and dynamic interplay between structure and texture. Specifically, we first design a structure enhancement module to restore the Histograms of Oriented Gradient (HOG) and the edge of an input image in a sketch space, which forms the input of the structure target stream. Meanwhile, in the image completion stream, we design a channel-space parallel-attention component to facilitate the efficient co-learning of channel and spatial visual cues. To build a bridge between the two streams, we further develop a structure-texture cross-attention module, wherein both structure and texture are synchronously extracted through self-attention, and texture extraction is implemented by dynamically tracking the structure in a cross-attention fashion, enabling the capture of the intricate interaction between structure and texture. Extensive experiments evaluated on three benchmark datasets, including CelebA, Places2, and Paris StreetView, demonstrate that the proposed TSFormer achieves state-of-the-art performance compared to its competitors. The code is available at https://github.com/GZHU-DVL/TSFormer .

Online Boosting Adaptive Learning under Concept Drift for Multistream Classification

Multistream classification poses significant challenges due to the necessity for rapid adaptation in dynamic streaming processes with concept drift. Despite the growing research outcomes in this area, there has been a notable oversight regarding the temporal dynamic relationships between these streams, leading to the issue of negative transfer arising from irrelevant data. In this paper, we propose a novel Online Boosting Adaptive Learning (OBAL) method that effectively addresses this limitation by adaptively learning the dynamic correlation among different streams. Specifically, OBAL operates in a dual-phase mechanism, in the first of which we design an Adaptive COvariate Shift Adaptation (AdaCOSA) algorithm to construct an initialized ensemble model using archived data from various source streams, thus mitigating the covariate shift while learning the dynamic correlations via an adaptive re-weighting strategy. During the online process, we employ a Gaussian Mixture Model-based weighting mechanism, which is seamlessly integrated with the acquired correlations via AdaCOSA to effectively handle asynchronous drift. This approach significantly improves the predictive performance and stability of the target stream. We conduct comprehensive experiments on several synthetic and real-world data streams, encompassing various drifting scenarios and types. The results clearly demonstrate that OBAL achieves remarkable advancements in addressing multistream classification problems by effectively leveraging positive knowledge derived from multiple sources.

Neural mechanisms of spatial auditory attention with magnified interaural level difference cues

Bilateral cochlear implant users struggle in spatial release from masking (SRM) tasks, likely due to restricted access to interaural time difference (ITD) cues. Instead, they must rely on interaural level difference (ILD) cues; however, our previous behavioral experiments suggest that magnification of ILDs can facilitate SRM. Here, we probed the neural mechanisms underlying the benefit of magnified ILDs. We tested 18 normal-hearing subjects in an anechoic chamber. Listeners heard target and masker sequences of object and color words from opposite (left and right) quarterfields and were asked to detect color words in the target stream. Both streams were spatialized using either 50 μS ITD (ITD50), 500 μS ITD (ITD500), a broadband 10 dB ILD (ILD10), or the largest naturally occurring frequency-specific ILD (70 degrees; ILD70n). We recorded task-elicited hemodynamic responses in dorsolateral prefrontal cortex (DLPFC) using functional Near-Infrared Spectroscopy. Subjects performed best in the ITD500 and ILD10 conditions. Hemodynamic response magnitudes were smaller for ITD50 than for all other conditions, consistent with frontal activity increasing when perceptual segregation is possible and spatial attention can be deployed successfully. These data show that magnified ILD cues enhance SRM and that the benefit ILDs confer arises because listeners can engage cognitive attentional processes.

The Relationship of Pitch Discrimination with Segregation of Tonal and Speech Streams for Cochlear Implant Users.

Cochlear implant (CI) users often complain about music appreciation and speech recognition in background noise, which depend on segregating sound sources into perceptual streams. The present study examined relationships between frequency and fundamental frequency (F0) discrimination with stream segregation of tonal and speech streams for CI users and peers with no known hearing loss. Frequency and F0 discrimination were measured for 1,000 Hz pure tones and 110 Hz complex tones, respectively. Stream segregation was measured for pure and complex tones using a lead/lag delay detection task. Spondee word identification was measured in competing speech with high levels of informational masking that required listeners to use F0 to segregate speech. The hypotheses were that frequency and F0 discrimination would explain a significant portion of the variance in outcomes for tonal segregation and speech reception. On average, CI users received a large benefit for stream segregation of tonal streams when either the frequency or F0 of the competing stream was shifted relative to the target stream. A linear relationship accounted for 42% of the covariance between measures of stream segregation and complex tone discrimination for CI users. In contrast, such benefits were absent when the F0 of the competing speech was shifted relative to the target speech. The large benefit observed for tonal streams is promising for music listening if it transfers to separating instruments within a song; however, the lack of benefit for speech suggests separate mechanisms, or special requirements, for speech processing.

Heterogeneous Domain Adaptation for Multistream Classification on Cyber Threat Data

Under a newly introduced setting of multistream classification, two data streams are involved, which are referred to as source and target streams. The source stream continuously generates data instances from a certain domain with labels, while the target stream does the same task without labels from another domain. Existing approaches assume that domains for both data streams are identical, which is not quite true, since data streams from different sources may contain distinct features. Indeed, they may even have different numbers of features. Furthermore, obtaining labels for every instance in a data stream is often expensive and time-consuming. Therefore, it has become an important topic to explore if classes of labeled instances from other related streams are helpful to predict the classes of unlabeled instances in a different stream. Note that domains of source and target streams may have distinct feature spaces and data distributions. Our objective is to predict class labels of data instances in the target stream by using the classifiers trained by the source stream. We propose a framework of multistream classification by using projected data from a common latent feature space, which is embedded from both source and target domains. This framework is also crucial for enterprise system defenders to detect cross-platform attacks, such as Advanced Persistent Threats (APTs). Empirical valuation and analysis on both real-world and synthetic datasets are performed to validate the effectiveness of our proposed algorithm, comparing to state-of-the-art techniques. Experimental results show that our approach significantly outperforms other existing approaches.

Autonomous Cross Domain Adaptation Under Extreme Label Scarcity.

A cross domain multistream classification is a challenging problem calling for fast domain adaptations to handle different but related streams in never-ending and rapidly changing environments. Notwithstanding that existing multistream classifiers assume no labeled samples in the target stream, they still incur expensive labeling costs since they require fully labeled samples of the source stream. This article aims to attack the problem of extreme label shortage in the cross domain multistream classification problems where only very few labeled samples of the source stream are provided before process runs. Our solution, namely, Learning Streaming Process from Partial Ground Truth (LEOPARD), is built upon a flexible deep clustering network where its hidden nodes, layers, and clusters are added and removed dynamically with respect to varying data distributions. A deep clustering strategy is underpinned by a simultaneous feature learning and clustering technique leading to clustering-friendly latent spaces. A domain adaptation strategy relies on the adversarial domain adaptation technique where a feature extractor is trained to fool a domain classifier by classifying source and target streams. Our numerical study demonstrates the efficacy of LEOPARD where it delivers improved performances compared to prominent algorithms in 15 of 24 cases. Source codes of LEOPARD are shared in https://github.com/wengweng001/LEOPARD.git to enable further study.

Energy, environmental, and economic analysis of a new configuration multi-stage flash distillation unit coupled with steam power plant

This study focuses on the utilization of three waste streams from a power plant for the provision of cooling water and required steam for heating purposes in a multi-stage flash desalination (MSF) unit. The mathematical model for a new configuration of MSF desalination is presented. In conventional MSF desalination unit, a stream of salt water is used for desalination and system coolant. A new structure for the desired desalination process, proportional to the salt concentration of the feed streams and based-on un-mixing of the target streams has been proposed. The proposed system required 18 m2 of cross-sectional area and also 87 m2/(kg/s) of specific heat transfer area in the power plant with seven stages. The optimal value of top brine temperature (TBT) has 110 °C and the capacity of the proposed MSF desalination unit is 92.43 m3 of fresh water/day. The gain output ratio (GOR), energy intensity and water cost of the system are calculated to be 3.12, 706.8 kJ/kg, and 0.69 $/m3, respectively. By application of proposed MSF desalination, the annual emission of 1167.5 tons CO2 into the atmosphere can be obviated and the cost savings resulting from equal fuel consumption amount to $116,315.5 per year.

Reduced-Space Multistream Classification Based on Multiobjective Evolutionary Optimization

In traditional data stream mining, classification models are typically trained on labeled samples from a single source. However, in real-world scenarios, obtaining accurate labels is very hard and expensive, especially when multiple data streams are concurrently sampled from an environment or the same process. To address this issue, multistream classification is proposed, in which a data stream with biased labels (called the source stream) is leveraged to train a suitable model for prediction over another stream with unlabeled samples (called the target stream). Despite the growing research in this field, previous multistream classification methods are mostly designed for single source stream scenarios. However, various source streams contain diverse data distributions, providing more valuable information for building a more accurate model. In addition, previous works construct classification models in the original shared feature space, ignoring the effect of redundant or low-quality features on the classification performance. This may produce inefficient knowledge transfer across streams. In view of this, a reduced-space multistream classification based on multi-objective evolutionary optimization is proposed in this paper. First, a multi-objective evolutionary optimization is employed to seek the most valuable feature subset shared in the source and target domains, with the purpose of narrowing the distribution difference between source and target streams. Following that, a Gaussian Mixture Model-based weighting mechanism for source samples is presented. More especially, two drift adaptation methods are proposed to address asynchronous drift. Experimental results on benchmark datasets show that the proposed method outperforms other comparative methods on classification accuracy and G-mean.

CSAL: Cost sensitive active learning for multi-source drifting stream

Multi-source stream classification is a prominent real-world problem challenged by the limited real labels and non-stationary environment. Despite growing research achievements in this field, most existing works solved this problem by requiring all real labels of source or target streams to conduct domain adaption or transfer learning mechanisms, which brings high labeling costs. However, in real-world applications, there are usually insufficient labeled data in both source and target streams, and the annotation cost of the source and target streams are generally unequal. Thus, we propose a Cost-Sensitive Active Learning (CSAL) method for multi-source drifting streams. Specifically, a multi-source ensemble framework with an asymmetry weighting mechanism is presented to ensure beneficial knowledge transfer and avoid the negative transfer. Then, a multi-perspective similarity estimation method is proposed to evaluate the similarity of source and target streams. On this basis, a novel cost-sensitive hybrid labeling strategy that combines volatility strategy and uncertainty strategy with a cost-sensitive budget control mechanism is proposed, which adaptively selects representative samples at the appropriate time. At last, a parallel multiple hypothesis drift detection method is proposed, which can efficiently utilize real labels to detect concept drift. Experimental results on real-world and synthetic data streams show that our CSAL outperforms the state-of-the-art methods with even fewer labels.

Mining Discriminative Itemsets Over Data Streams Using Efficient Sliding Window

In this paper, we present an efficient novel method for mining discriminative itemsets over data streams using the sliding window model. Discriminative itemsets are the itemsets that are frequent in the target data stream, and their frequency in the target stream is much higher in comparison to their frequency in the rest of the streams. The problem of mining discriminative itemsets has more challenges than mining frequent itemsets, especially in the sliding window model, as during the window frame sliding, the algorithms have to deal with the combinatorial explosion of itemsets in more than one data stream, for the transactions coming in and going out of the sliding window. We propose a single scan algorithm using two novel in-memory data structures for mining discriminative itemsets in a combination of offline and online sliding windows. Offline processing is used for controlling the generation of many unpromising itemsets. Online processing is used for getting more up-to-date and accurate online answers between two offline slidings. The discovered discriminative itemsets are accurately updated in the offline sliding window periodically, and the mining process is continued in the online sliding between two periodic offline slidings. The extensive empirical analysis shows that the proposed algorithm provides efficient time and space complexities with full accuracy. The algorithm can handle large, fast-speed, and complex data streams.

Hybrid membrane process for water treatment: a short review

Abstract Water shortage is one of the most difficult issues confronting people all over the world. Rapid urbanization and water scarcity necessitate immediate action to improve sustainable water management without jeopardizing global socioeconomic growth. Thus, conventional water treatments are implemented for the purpose of eradicating various pollutants in wastewater. Traditional water treatment methods, whether in water treatment facilities or reverse osmosis (RO) plants, have run across a number of roadblocks that have significantly hampered their performance and efficiency. Integrating the membrane process with other remediation technologies in a hybrid process is a novel technique to improve contaminant extraction efficiency for our target streams. This process is termed the hybrid membrane process (HMP). On this aspect, this paper would highlight the benefits of using the HMP compared to conventional methodologies and their applications conducted in various sectors around the world. Some case studies are also reviewed illustrating its cost analysis in comparison to conventional methodologies accentuating the merits of using HMPs.

Extending Height Above Nearest Drainage to Model Multiple Fluvial Sources in Flood Inundation Mapping Applications for the U.S. National Water Model

AbstractHeight Above Nearest Drainage (HAND), a drainage normalizing terrain index, is a means able of producing flood inundation maps (FIMs) from the National Water Model (NWM) at large scales and high resolutions using reach‐averaged synthetic rating curves. We highlight here that HAND is limited to producing inundation only when sourced from its nearest flowpath, thus lacks the ability to source inundation from multiple fluvial sources. A version of HAND, known as Generalized Mainstems (GMS), is proposed that discretizes a target stream network into segments of unit Horton‐Strahler stream order known as level paths (LPs). The FIMs associated with each independent LP are then mosaiced together, effectively turning the stream network into discrete groups of homogeneous unit stream order by removing the influence of neighboring tributaries. Improvement in mapping skill is observed by significantly reducing false negatives at river junctions when the inundation extents are compared to FIMs from that of benchmarks. A more marginal reduction in the false alarm rate is also observed due to a shift introduced in the stage‐discharge relationship by increasing the size of the catchments. We observe that the improvement of this method applied at 4%–5% of the entire stream network to 100% of the network is about the same magnitude improvement as going from no drainage order reduction to 4%–5% of the network. This novel contribution is framed in a new open‐source implementation that utilizes the latest combination of hydro‐conditioning techniques to enforce drainage and counter limitations in the input data.

Defining functional spatial boundaries using a spatial release from masking task.

The classic spatial release from masking (SRM) task measures speech recognition thresholds for discrete separation angles between a target and masker. Alternatively, this study used a modified SRM task that adaptively measured the spatial-separation angle needed between a continuous male target stream (speech with digits) and two female masker streams to achieve a specific SRM. On average, 20 young normal-hearing listeners needed less spatial separation for 6 dB release than 9 dB release, and the presence of background babble reduced across-listener variability on the paradigm. Future work is needed to better understand the psychometric properties of this adaptive procedure.

Do we parse the background into separate streams in the cocktail party?

In the cocktail party situation, people with normal hearing usually follow a single speaker among multiple concurrent ones. However, there is no agreement in the literature as to whether the background is segregated into multiple streams/speakers. The current study varied the number of concurrent speech streams and investigated target detection and memory for the contents of a target stream as well as the processing of distractors. A male-voiced target stream was either presented alone (single-speech), together with one male-voiced distractor (one-distractor), or a male- and a female-voiced distractor (two-distractor). Behavioral measures of target detection and content tracking performance as well as target- and distractor detection related event-related brain potentials (ERPs) were assessed. We found that the N2 amplitude decreased whereas the P3 amplitude increased from the single-speech to the concurrent speech streams conditions. Importantly, the behavioral effect of distractors differed between the conditions with one vs. two distractor speech streams and the non-zero voltages in the N2 time window for distractor numerals and in the P3 time window for syntactic violations appearing in the non-target speech stream significantly differed between the one- and two-distractor conditions for the same (male) speaker. These results support the notion that the two background speech streams are segregated, as they show that distractors and syntactic violations appearing in the non-target streams are processed even when two speech non-target speech streams are delivered together with the target stream.

Automatic damage segmentation in pavement videos by fusing similar feature extraction siamese network (SFE-SNet) and pavement damage segmentation capsule network (PDS-CapsNet)

Discovering damage in pavement videos is critical for the assessment of pavement conditions. Unfortunately, existing damage segmentation methods in pavement videos have problems of low efficiency, low accuracy, and difficulty in adapting to changes in the appearance of the damage. An automatic damage segmentation system for pavement videos is proposed. The system is formed of a similar feature extraction Siamese network (SFE-SNet) and a pavement damage segmentation Capsule Network called PDS-CapsNet. First, SFE-SNet is developed based on an improved Siamese network modified by a dynamically updating reference frame strategy to extract similar features of the target stream and the reference stream in pavement videos. The difficulty of adapting to changes in the appearance of damage is solved by the dynamic updating of the reference frame strategy. Second, for the first time, PDS-CapsNet is developed based on an improved CapsNet modified by using the padding convolution method, adding an upsampling layer and adding the target stream low-order feature strategy to segment damage in pavement videos. Then, two datasets containing 2386 images of pavement defects and 216 pavement videos are created for pre-training and precise training. The trained model achieves the Jaccard coefficient of 96.87% and F1-score of 95.79% on the validation set. Compared with five other state-of-the-art methods, the F1-score of the proposed model improved by 5.32%, 9.81%, 3.44%, 10.31%, and 8.99%, respectively. Subsequently, the proposed system is embedded in a pavement damage detection vehicle to realize the display of damage segmentation results in real time. An on-site experiment is carried out to verify the validity and accuracy of the system.

ACDC: Online unsupervised cross-domain adaptation

We consider the problem of online unsupervised cross-domain adaptation, where two independent but related data streams with different feature spaces – a fully labeled source stream and an unlabeled target stream – are learned together. Unique characteristics and challenges such as covariate shift, asynchronous concept drifts, and contrasting data throughput arise. We propose ACDC, an adversarial unsupervised domain adaptation framework that handles multiple data streams with a complete self-evolving neural network structure that reacts to these defiances. ACDC encapsulates three modules into a single model: A denoising autoencoder that extracts features, an adversarial module that performs domain conversion, and an estimator that learns the source stream and predicts the target stream. ACDC is a flexible and expandable framework with little hyper-parameter tunability. Our experimental results under the prequential test-then-train protocol indicate an improvement in target accuracy over the baseline methods, achieving more than a 10% increase in some cases.

High-rate conversion of lactic acid-rich streams to caproic acid in a fermentative granular system

Lactic acid-driven chain elongation enables upgrading low-value organic streams into caproic acid. Recently, volumetric production rates over 0.5 g L-1 h-1have been reported for carbohydrate-rich streams in expanded granular sludge bed (EGSB) reactors. However, many target streams contain mixtures of carbohydrates and lactic acid, and little is known about their impact on product profile and microbial ecology, or the importance of carbohydrates as substrate to achieve high rates. This manuscript investigated varying glucose-to-lactate ratios and observed that decreasing glucose-content eliminated odd-chain by-products, while glucose omission required acetic acid addition to support lactic acid conversion. Decreasing the glucose-content fed resulted in decreasing amounts of granular biomass, with the disappearance of granules when no glucose was fed. Lowering the HRT to 0.3 days while feeding only lactic and acetic acid likely triggered re-granulation, enabling the highest lactic acid-driven caproic acid production rates reported thus far at 16.4 ± 1.7 g L-1 d-1.

Cat-astrophic effects of sudden interruptions on spatial auditory attention.

Salient interruptions draw attention involuntarily. Here, we explored whether this effect depends on the spatial and temporal relationships between a target stream and interrupter. In a series of online experiments, listeners focused spatial attention on a target stream of spoken syllables in the presence of an otherwise identical distractor stream from the opposite hemifield. On some random trials, an interrupter (a cat "MEOW") occurred. Experiment 1 established that the interrupter, which occurred randomly in 25% of the trials in the hemifield opposite the target, degraded target recall. Moreover, a majority of participants exhibited this degradation for the first target syllable, which finished before the interrupter began. Experiment 2 showed that the effect of an interrupter was similar whether it occurred in the opposite or the same hemifield as the target. Experiment 3 found that the interrupter degraded performance slightly if it occurred before the target stream began but had no effect if it began after the target stream ended. Experiment 4 showed decreased interruption effects when the interruption frequency increased (50% of the trials). These results demonstrate that a salient interrupter disrupts recall of a target stream, regardless of its direction, especially if it occurs during a target stream.

A novel paradigm to define functional spatial boundaries

One approach to studying an individual’s binaural function is to measure speech reception ability for spatially separated compared to co-located competing speech streams. This spatial-release-from-masking (SRM) task has been widely adopted due to its capacity to detect abnormal binaural function, but it has its limitations in interpreting individual spatial abilities. Here, we introduce a modified SRM task that directly measures the needed spatial separation between target and maskers to achieve a specific SRM. Stimuli were presented from 24 speakers in the free field using an equal power panning method. Normal-hearing listeners (n = 10) were tasked with identifying digits in a target stream of monosyllabic words always presented from the front. Initial measures were made using the classic SRM paradigm in which two opposite-gender masker streams were either presented from the front (co-located), or symmetrically from 30° (fixed-separation), to determine a speech reception threshold (SRT; in dB target-to-masker ratio). Based on the co-located SRT, the task then adaptively tracked the separation angle needed to achieve 6-dB SRM. Results showed for most listeners, a 10-dB SRM at 30° separation and an average spatial boundary less than 30 degrees when fixed at 6-dB SRM. [Work supported by NIH NIDCD R21DC017832 to EJO.]